EP0559377A2

EP0559377A2 - Noise shaping circuit

Info

Publication number: EP0559377A2
Application number: EP93301410A
Authority: EP
Inventors: Shinichirou c/o Sony Corporation Miyazaki; Akira c/o Sony Corporation Shirahama
Original assignee: Sony Corp
Current assignee: Sony Corp
Priority date: 1992-02-29
Filing date: 1993-02-25
Publication date: 1993-09-08
Also published as: JPH05244011A; EP0559377A3; US5396248A

Abstract

A noise shaping circuit is disclosed in which the second significant bit of quantization input data in 2's complement representation, supplied to a quantizer, is taken out after complementation by an inverter 30, while the second significant bit and following bit or bits are taken out as such without complementation, for producing a quantization error, which is subtracted from an input signal to render it possible to abbreviate an operation of finding the quantization error. A sole inverter may be used in substitution for a circuit portion employed for finding the quantization error for significantly diminishing the circuit scale.

Description

This invention relates to a noise shaping circuit for reducing the quantization error produced in a quantizer for audio signals.
Heretofore, when sampling input signals at a sampling frequency fs, quantizing the sampled input signals into digital signals and requantizing the digital signals for reducing the number of bits, quantization noises produced at the re-quantizer are frequently fed back via a noise filter to an input side of the quantizer to reduce the quantization noise or quantization distortion by way of performing a quantization error reducing operation by error feedback referred to hereinafter as noise shaping. It is the noise shaping circuit which carries out the noise shaping. With the noise shaping circuit, the spectrum of the quantization noise is moved to a frequency range outside the audible range, for example, to a higher frequency range outside the audible range, for improving the S/N ratio within the audible range.
In general, in a first order noise shaping circuit of a basic construction, the quantization noise is found from a difference between an input and an output of a quantization circuit. However, when carrying out this operation of finding the difference, a large number of adders and flip-flops are required in a practical system for inputting data of, for example the order of several bits, such that the adders and the flip- flops account for about 30% of the first order noise shaping circuit, thus enlarging the circuit scale.
In view of the above-described status of the art, it is an object of the present invention to provide a noise shaping circuit whereby the circuit scale for carrying out an operation of the quantization noise may be diminished.
According to the present invention, there is provided a noise shaping circuit in which a quantization error component in a quantizer for quantizing input signals is fed back via a predetermined feedback circuit to an input of said quantizer, wherein
when taking out said quantization error, the second bit as counted from the most significant bit of input quantization data supplied to said quantizer is complemented.
It should be noticed that the second significant bit of the quantization input data in 2's complement representation corresponds to the most significant bit of the input data supplied to the noise shaping circuit.
With the noise shaping circuit according to the present invention, quantization error outputting means output the second significant bit of quantization input data in 2's complement representation supplied to a quantizer after complementation, and output the second significant bit and any following bit or bits of the quantization input data without complementation, while a subtracter subtracts the quantization error from the input data supplied to the noise shaping circuit, to render it possible to abbreviate an operation of finding the quantization error. A sole inverter may be used in substitution for a circuit portion for finding the quantization error, made up of full adders and D-flipflops provided with a large number of gates to render it possible to reduce the circuit scale significantly.
The invention will be further described by way of non-limitative example with reference to the accompanying drawings, in which:-
Fig.1 is a functional block circuit diagram showing the construction of a basic first order noise shaping circuit according to the present invention.
Fig.2 is schematic view showing an arrangement of a 4 input bit first order noise shaping circuit derived from the basic first order noise shaping circuit shown in Fig.1.
Fig.3 is a schematic view showing an arrangement of a 4 input bit first order noise shaping circuit to which an embodiment of a noise shaping circuit according to the present invention is applied.
Before proceeding to describing the preferred embodiment of the present invention, the construction of the basic first order noise shaping circuit is explained by referring to Fig.1.
In the noise shaping circuit, shown in Fig.1, audio signals supplied to an input terminal 61 are supplied via an additive node 51 to a quantization circuit 52 where they are quantized into digital audio signals of predetermined numbers of bits. The noise shaping circuit effects noise shaping of the quantization noise generated by the quantization circuit 52. That is, the signal supplied to the quantization circuit 52 is subtracted at an additive node 53 from a quantization output of the quantization circuit 52 to produce a quantization noise which is supplied to a delay circuit 54. An output of the delay circuit 54 is supplied to the additive node 51. The delay circuit 54 delays the quantization noise by a predetermined time duration corresponding to one sample Z⁻¹. Consequently, with the above-described noise shaping circuit, the output signal delayed by one sample is subtracted from the input of the adder 51.
The noise shaping circuit shown in Fig.2 is a 4 input bit first order noise shaping circuit constructed on the basis of the basic first order noise shaping circuit shown in Fig.1.
The 4 input bit first order noise shaping circuit shown in Fig.2 is made up of a first full adder section 105, composed of full adders 100, 101, 102, 103 and 104, a second full adder section 115 composed of full adders 110, 111, 112, 113 and 114, a latch section 125 composed of data flipflops or D- flipflops 120, 121, 122 ,123 and 124 and an inverter 130 connected across the full adder 104 of the first full adder 105 and the full adder 114 of the full adder section 115.
Within the first full adder section 105, a carry output terminal C_O0 of the full adder 100 is connected to carry input terminal C_I1 of the upper full adder 101, a carry output terminal C_O1 of which is connected to a carry input terminal C_I2 of the upper full adder 102. The same state of connection as the above- described state of connection between the carry output terminal and the carry input terminal of the upper order full adder is maintained between the similar terminals of the full adders 102 and 103. However, a carry output terminal C_O4 of the full adder 104 is open-circuited, whereas a carry input terminal C_I0 of the full adder 100 is grounded and is permanently at a low (L) level.
Within the second full adder section 115, a carry output terminal C_O₀ of the full adder 110 is connected to a carry input terminal C _I₁ of the upper order full adder 111, a carry output terminal C _O₁ of which is connected to a carry input terminal C_I2 of the upper order full adder 112. The same state of connection as the above-described state of connection between the carry output terminal and the carry input terminal of the upper order full adder is maintained between the similar terminals of the full adders 112 113 and 114. However, a carry output terminal C _O₄ of the full adder 114 is open-circuited, whereas a carry input terminal C_I0 of the full adder 110 is grounded and is permanently at a low (L) level. Input terminals a₀, a₁ and a₂ of the full adders 110, 111 and 112 are connected in common and grounded.
Within the latch section 125, clock output terminals C₀, CK₁, CK₂, CK₃ and CK₄ of the D- flipflops 120, 121, 122, 123 and 124 are connected in common and fed with clocks ∅.
As for connection between the full adders 100, 110, D- flipflop 120 and a data input terminal 130, the data input terminal 130 is connected to input terminal A₀ of full adder 100, an addition output terminal S₀ of which is connected to an input terminal b₀ of the full adder 110, an addition output terminal S₀ of which is connected to a data input terminal D₀ of D-flipflop 120, an affirmation output terminal Q₀ of which is connected to an input terminal B₀ of the full adder 100.
As for connection between the full adders 101, 111, D- flipflop 121 and a data input terminal 131, data input terminal 131 is connected to input terminal A₁ of full adder 101, an addition output terminal S₁ of which is connected to an input terminal b₁ of the full adder 111, an addition output terminal S₁ of which is connected to a data input terminal D₁ of D-flipflop 121, an affirmation output terminal Q₁ of which is connected to an input terminal B₁ of the full adder 101.
As for connection between the full adders 102, 112, D- flipflop 122 and a data input terminal 132, data input terminal 132 is connected to an input terminal A₂ of the full adder 102, an addition output terminal S₂ of which is connected to an input terminal b₂ of the full adder 112, an addition output terminal S₂ of which is connected to a data input terminal D₂ of D-flipflop 122, an affirmation output terminal Q₂ of which is connected to an input terminal B₂ of the full adder 102.
As for connection between the full adders 103, 113, D- flipflop 123 and a data input terminal 133, the data input terminal 133 is connected to an input terminal A₃ of the full adder 103, an addition output terminal S₃ of which is connected to an input terminal b₃ of the full adder 113, an addition output terminal S₃ of which is connected to a data input terminal D₃ of D-flipflop 123, an affirmation output terminal Q₃ of which is connected to an input terminal B₃ of the full adder 103.
As for connection between the full adders 104, 114, D- flipflop 124 and a data input terminal 133, the data input terminal 133 is connected to an input terminal A₄ of the full adder 104, an addition output terminal S₄ of which is connected to an input terminal b₄ of the full adder 114, an addition output terminal S₄ of which is connected to a data input terminal D₄ of D-flipflop 124, an affirmation output terminal Q₄ of which is connected to an input terminal B₄ of the full adder 104.
An inverter 130 is connected across a jumper connecting the addition output terminal S₄ of the full adder 104 with the input terminal b₄ of the full adder 114 and the input terminal a₄ of the full adder 114.
Input data D₁₀, D₁₁, D₁₂ and D₁₃, associated with 2⁻³, 2⁻², 2⁻¹ and 20 digits, respectively, are fed to input terminals 130, 131, 132 and 133, respectively.
Meanwhile, if, in the basic first order noise shaping circuit shown in Fig.1, an input supplied from input terminal 61 is x(n), an output from an output terminal 62 is y₀(n) an input to quantization circuit 52 is y₁(n) and an output from delay circuit 54 is y2 (n), the following formulas $y2 (n + 1) = y₁(n) - y₀(n)$
$y₁(n) = x(n) + y2(n)$
$If y₁(n) ≧ 0, y₀(n) = 1$
$If y₁(n) < 0, y₀(n) = -1$
Meanwhile, the value of "1" for y₀(n) when y₁(n) ≧ 0 (formula 3) and the value of "1" for y₀(n) when y₁(n) < 0 (formula 4), in 2's complement representation in 5 bits, are as shown in Table 1: TABLE 1

y₀(n) sign bit 2⁰ 2⁻¹ 2⁻² 2⁻³

1 0 1 0 0 0

-1 1 1 0 0 0

The most significant bits (MSBs) of the two words shown in Table 1 are sign bits, with "0" and "1" indicating positive and negative signs, respectively.
If y₁(n) = 0, y₀(n) = 1 holds from formula (3), so that the right side of the formula (1) becomes $y₁(n) - 1$
The operation of the formula (5), performed in the above- mentioned 2's complement representation in 5 bits, is as shown in the following formula (6) :
In formula (6), the MSB of the 5-bit word for y₁(n) and other bits are set to 0 and x, respectively, where x is equal to 0 or 1 (x = 0 or 1).
Since the lower three bits "000" of the 5-bit word in 2's complement representation for "+1" as an addition input are added to the lower three bits "xxx" for y₁(n) of formula (6), the lower three bits of the result of addition remain to be "xxx". Since it is the second significant bit "1" of the addition input that is added to the second significant bit of y₁(n), the second significant bit of the result of addition is the complemented second significant bit "x" of y₁(n), or "x*". Since the MSB of the result of addition differs depending on whether the second significant bit "x" of y₁(n) is "0" or "1", it is represented as "?". However, since the result of y₁(n) - 1 of the formula (5) in necessarily in a range of ±1, the result of addition is represented by four bits. If the second significant bit of y₁(n) is "0", that is the digit 2⁰ is 0, y₁(n) - 1 is less than 0, that is, y₁(n) - 1 < 0. That is "x*" of the result of addition is "1", such that the MSB or the sign bit of the results of addition becomes "1", with the result of addition becoming negative. On the other hand, if the second significant bit of y₁(n) is "1", that is the digit 2⁰ is 1, y₁(n) ≧ 1 so that (y₁(n) - 1 ≧ 0. Consequently, "x*" of the result of addition becomes "0" because of the carry output. Since the carry input "1" is added to the MSB which is the sign bit of the result of addition, the sign bit becomes 0''. That is, the result of addition becomes positive. In other words, an operation of y₁(n) - 1 for obtaining the result of addition has been performed by complementing the second significant bit of y₁(n), with the lower three bits remaining unchanged.
Besides, if y₁(n) < 0, y₀(n) = -1 from formula 4, so that the right side of the formula (1) becomes $Y₁(n) + 1$
The operation of the formula 7 is performed in the form of the above-mentioned 2's complement representation. The following formula (8)

then is obtained.
Since the lower three bits "000" of the 5-bit word "+1" as an addition input in 2's complement representation are added to the lower three bits "xxx" for y₁(n) of formula (8), the lower three bits of the result of addition remain to be "xxx". Since it is the second significant bit "1" of the addition input that is added to the second significant bit of y₁(n), the second significant bit of the result of addition is the complemented second significant bit "x" of y₁(n), or "x*". Since the MSB of the result of addition differs depending on whether the second significant bit "x" of y₁(n) is "0" or "1", it is represented as "?". Since the result of y₁(n) - 1 of the formula (7) is necessarily in a range of ±1, the result of addition is represented by four bits. If the second significant bit of y₁(n) is "1", that is the digit 2⁰ is 1, y₁(n) ≧ -1, that is, y₁(n) + 1 of formula 7 is not less than 0 (y₁(n) + 1 ≧ 0). That is "x*" of the result of addition is "0", because of a carry input, Thus the MSB or the sign bit of the results of addition becomes "0", because of the carry input, with the result of addition becoming positive. On the other hand, if the second significant bit of y₁(n) is "0", that is if the digit 2⁰ is 0, y₁(n) ≧ 1, so that (y₁(n) - 1 ≧ 0. Consequently, y₁(n) + 1 of formula 7 is less than 0 (y₁(n) + 1 < 0). Therefore, the "x*" of the result of addition becomes "1" and the MSB which is the sign bit of the result of addition becomes "1". That is, the result of addition becomes positive. In other words, an operation of y₁(n) + 1 for obtaining the result of addition has been performed by complementing the second significant bit of y₁(n), with the lower three bits remaining unchanged.
It will be seen from above that the operation of Y₁(n) - Y₀(n), which is the right side of the formula (1), is within a range of ±1 and may be performed in any case by deleting the MSB of y₁(n) and complementing the second significant bit of y₁(n). This indicates that the 4 input bit first order noise shaping circuit may be implemented by the arrangement shown in Fig.3.
Referring to Fig.3, the 4 input bit noise shaping circuit, applying the noise shaping circuit according to the present invention, is made up of a full adder section 15 composed of full adders 10, 11, 12, 13 and 14, a latch section 25 composed of D- flipflops 20 ,21, 22 and 23, and an inverter 30 connected across the full adder 13 and the D-flipflop 23.
Within the second full adder section 15, a carry output terminal C_o0 of the full adder 10 is connected to a carry input terminal C_i1 of the upper full adder 11, a carry output terminal C_o1 of which is connected to a carry input terminal C_i2 of the upper full adder 12. The same state of connection as the above- described state of connection between the carry output terminal of the lower full adder and the carry input terminal of the upper full adder is maintained between the similar terminals of the full adders 12, 13 and 14. However, a carry output terminal C_o4 of the full adder 14 is open-circuited, whereas a carry input terminal C_i0 of the full adder 10 is grounded and is perpetually at a low (L) level.
Within the latch section 25, clock terminals ck₀, ck₁, ck₂ and ck₃ of the D- flipflops 20, 21, 22 and 23 are connected in common and to a jumper from an affirmation output terminal Q₀ of the D-flipflop 20 to an input terminal B₀ of the full adder 10.
As for connection between the full adder 10, D-flipflop 20 and a data input terminal 50, the data input terminal 50 is connected to input terminal A₀ of full adder 10, an addition output terminal S₀ of which is connected to a data input terminal D₀ of D-flipflop 20, an affirmation output terminal Q₀ of which is connected to an input terminal B₀ of the full adder 10.
As for connection between the full adder 11, D-flipflop 21 and a data input terminal 51, the data input terminal 51 is connected to input terminal A₁ of full adder 11, an addition output terminal S₁ of which is connected to a data input terminal D₁ of D-flipflop 21, an affirmation output terminal Q₁ of which is connected to an input terminal B₁ of the full adder 11.
As for connection between the full adder 12, D-flipflop 22 and a data input terminal 52, the data input terminal 52 is connected to input terminal A₂ of full adder 12, an addition output terminal S₂ of which is connected to a data input terminal D₂ of D-flipflop 22, an affirmation output terminal Q₂ of which is connected to an input terminal B₂ of the full adder 12.
As for connection between the full adder 13, D-flipflop 23 and a data input terminal 53, the data input terminal 53 is connected to input terminal A₃ of full adder 13, an addition output terminal S₃ of which is connected to a data input terminal D₃ of D-flipflop 23, an affirmation output terminal Q₃ of which is connected to an input terminal B₃ of the full adder 13.
The full adder 14 has its input terminals A₄ and B₄ connected to data input terminal 53 and to input terminal B₃ of the full adder 13, while having its addition output terminal S₄ connected to output terminal 40.
Input data D_i0, D_i1, D_i2 and D_i3, associated with places 2⁻³, 2⁻², 2⁻¹ and 2⁰ respectively, are fed to input terminals 50, 51, 52 and 53, respectively.
As compared to the conventional 4 input bit primary noise shaping circuit shown in Fig.2, the above-described 4 input bit primary noise shaping circuit shown in Fig.3 is not provided with the five full adders 110, 111, 112, 113 and 114 of the full adder section 115, while it is also not provided with the D-flipflop 124 of the latch section 125. Also, as compared with the basic first order noise shaping circuit shown in Fig.1, the additive node 53 for effecting subtraction is eliminated.
Consequently, with the 4 input bit first order noise shaping circuit, applying an embodiment of the noise shaping circuit according to the present invention, the operation of the portion of the circuit constituted by the full adders and the D-flipflop each having a larger number of gates may be performed by a sole inverter to decrease the circuit scale significantly.
It is to be noted that the noise shaping circuit according to the present invention is not limited to the above-described embodiments. For example, the number of inputs to the noise shaping circuit may be any arbitrary number other then four.

Claims

A noise shaping circuit in which a quantization error component in a quantizer for quantizing input signals is fed back via a predetermined feedback circuit to an input of said quantizer, wherein
when taking out said quantization error, the second bit as counted from the most significant bit of input quantization data supplied to said quantizer is complemented.
A noise shaping circuit in which a quantization error component in a quantizer for quantizing input signals is fed back via a predetermined feedback circuit to an input of said quantizer, wherein
when taking out said quantization error, the second bit as counted from the most significant bit of input quantization data supplied to said quantizer is complemented, the third and subsequent bits of said input quantization data are taken out as such without complementation.
A noise shaping circuit in which a quantization error component in a quantizer for quantizing input signals is fed back via a predetermined feedback circuit to an input of said quantizer, comprising
quantization error outputting means for outputting a quantization error by taking out the second bit as counted from the most significant bit of quantized input data supplied to said quantizer after complementation and taking out the second and the following bits of said quantized input data as such without complementation, and
subtracting means for subtracting said
quantization error from said quantization error outputting means from said input signal.
A noise shaping circuit in which a quantization error component in a quantizer for quantizing input signals is fed back via a predetermined feedback circuit to an input of said quantizer, comprising quantization error outputting means for outputting a quantization error by taking out the second bit as counted from the most significant bit of quantized input data in 2's complement representation supplied to said quantizer after complementation and taking out the second and the following bits of said quantized input data as such without complementation, and subtracting means for subtracting said quantization error from said quantization error outputting means from said input signal.
A noise shaping circuit in which a quantization error component in a quantizer for quantizing input signals is fed back via a predetermined feedback circuit to an input of said quantizer, comprising
quantization error outputting means for taking out said quantization error so that the second bit as counted from the most significant bit of input quantization data supplied to said quantizer is complemented and the third and subsequent bits of said input quantization data are taken out as such without complementation, said input quantization data being data in which the most significant bit is a sign bit, the next following bit is a unit digit and the following bit or bits are subdecimal digit or digits, and
a subtracter for subtracting the quantization error of said quantization error outputting means from said input signals.
A noise shaping circuit in which a quantization error component in a quantizer for quantizing N-bit input signals is fed back via a predetermined feedback circuit to an input of said quantizer, comprising
a full adder section fed with said N-bit input signals, an inverter for complementing the second significant bit of an (N + 1) bit output signal of said full adder section, and
an N-bit latch section fed with an inverted output signal from said inverter and the third significant and the following bit or bits of said output signal of said full adder section,
N-bit output signals from said latch section being supplied to said full adder section so as to be added to the n-bit input signals, with the most significant bit of the (N + 1) bit addition output signals being taken out as an output signal.
A noise shaping circuit in which a quantization error component in a quantizer quantizing an input signal in a range of ±1 into a value of +1 or -1 is fed back via a predetermined feedback circuit section to an input of said quantizer, comprising
quantization error outputting means for outputting a quantization error by taking out the second bit as counted from the most significant bit of quantized input data in 2's complement representation supplied to said quantizer after complementation and taking out the second and the following bits of said quantized input data as such without complementation, and
subtracting means for subtracting said quantization error from said quantization error outputting means from said input signal.
A noise shaping circuit in which a quantization error component in a quantizer quantizing an input signal in a range of ±1 into a value of +1 or -1 is fed back via a predetermined feedback circuit section to an input of said quantizer, comprising
quantization error outputting means for taking out said quantization error so that the second bit as counted from the most significant bit of input quantization data supplied to said quantizer is complemented and the third and subsequent bits of said input quantization data are taken out as such without complementation, said input quantization data being data in which the most significant bit is a sign bit, the next following bit is a unit digit and the following bit or bits are subdecimal digit or digits, and
a subtracter for subtracting the quantization error of said quantization error outputting means from said input signals.
A noise shaping circuit in which a quantization error component in a quantizer quantizing an N-bit input signal having a value in a range of ±1 into a value of +1 or -1 is fed back via a predetermined feedback circuit section to an input of said quantizer, comprising
a full adder section fed with said N-bit input signals,
an inverter for complementing the second significant bit of an (N + 1) bit output signal of said full adder section, and
an N-bit latch section fed with an inverted output signal from said inverter and the third significant and the following bit or bits of said output signal of said full adder section,
N-bit output signals from said latch section being supplied to said full adder section so as to be added to the n-bit input signals, with the most significant bit of the (N + 1) bit addition output signals being taken out as an output signal.
A noise shaping circuit in which a quantization error component in a quantizer quantizing an N-bit input signal having a value in a range of ±1 into a value of +1 or -1 is fed back via a predetermined feedback circuit section to an input of said quantizer, comprising
a full adder section fed with said N-bit input signals, said full adder section being composed of (N + 1) full adders,
an inverter for complementing the second significant bit of an (N + 1) bit output signal of said full adder section, and
a latch section composed of N D-flipflops, said latch section being fed with an inverted output signal from said inverter and the third significant and the following bit or bits of said output signal of said full adder section,
N-bit output signals from said latch section being supplied to said full adder section so as to be added to the n-bit input signals, with the most significant bit of the (N + 1) bit addition output signals being taken out as an output signal.
A noise shaping circuit in which a quantization error component in a quantizer quantizing an input signal having a value in a range of ±1 into a value of +1 or -1 is fed back via a predetermined feedback circuit section to an input of said quantizer, comprising
a full adder section composed of N full adders fed with said N-bit input signals and a full adder of the highest order supplied with the most significant bit of said N-bit input signals,
an inverter for complementing the second significant bit of an (N + 1) bit output signal of said full adder section, and
a latch section composed of N flipflops, said latch section being fed with an inverted output signal from said inverter and the third significant and the following bit or bits of said output signal of said full adder section,
N-bit output signals from said latch section being supplied to said N full adders of said full adder section and the most significant bit of N-bit output signals of said latch section being supplied to said full adder for the most significant bit of said full adder section for addition to said N-bit input signals, with an output signal for the full adder for the most significant bit of said full adder section being taken out as a sign bit of the quantization output.
A noise shaping circuit in which a quantization error component in a quantizer quantizing an N-bit input signal having a value in a range of ±1 into a value of +1 or -1 is fed back via a predetermined feedback circuit section to an input of said quantizer, comprising
a full adder section composed of N full adders fed with said N-bit input signals and a highest order full adder supplied with the most significant bit of said N-bit input signals,
an inverter for complementing the second significant bit of an (N + 1) bit output signal of said full adder section, and
a latch section composed of N D-flipflops, said latch section being fed with an inverted output signal from said inverter and the third significant and the following bit or bits of said output signal of said full adder section,
N-bit output signals from said latch section being supplied to said N full adders of said full adder section and the most significant bit of N-bit output signals of said latch section being supplied to said full adder for the most significant bit of said full adder section for addition to said N-bit input signals, with an output signal for the full adder for the most significant bit of said full adder section being taken out as a sign bit of the quantization output.
A noise shaping circuit in which a quantization error component in a quantizer quantizing a 4-bit input signal having a value in a range of ±1 into a value of +1 or -1 is fed back via a predetermined feedback circuit section to an input of said quantizer, comprising
a full adder section being composed of 4 full adders fed with said 4-bit input signals and a highest order full adder supplied with the most significant bit of said 4-bit input signals,
an inverter for complementing the second significant bit of an 5 bit output signal of said full adder section, and
a latch section composed of 4 flipflops, said latch section being fed with an inverted output signal from said inverter and the third significant and the following bit or bits of said output signal of said full adder section,
4-bit output signals from said latch section being supplied to said 4 full adders of said full adder section and the most significant bit of 4-bit output signals of said latch section being supplied to said full adder for the most significant bit of said full adder section for addition to said 4-bit input signals, with an output signal for the most significant bit of said full adder section being taken out as a sign bit of the quantization output.
A noise shaping method for feeding back a quantization error component in a quantizer quantizing input signals to an input of said quantizer via a predetermined feedback circuit, comprising
complementing the second bit as counted from the most significant bit of quantization input data supplied to said quantizer.
A noise shaping method for feeding back a quantization error component in a quantizer quantizing input signals to an input of said quantizer via a predetermined feedback circuit, comprising
taking out the second significant bit of quantization input data supplied to said quantizer after complementation and taking out the second significant bit and the following bit or bits of said quantization input data as such without complementation for producing a quantization error.
A noise shaping method for feeding back a quantization error component in a quantizer quantizing input signals to an input of said quantizer via a predetermined feedback circuit, comprising the steps of
taking out the second significant bit of quantization input data supplied to said quantizer after complementation and taking out the second significant bit and the following bit or bits of said quantization input data as such without complementation for outputting a quantization error, and
subtracting the quantization error from the preceding step from said input signal.
A noise shaping method for feeding back a quantization error component in a quantizer quantizing input signals to an input of said quantizer via a predetermined feedback circuit, comprising the steps of
taking out the second significant bit of quantization input data in a 2's complement representation supplied to said quantizer after complementation and taking out the second significant bit and the following bit or bits of said quantization input data as such without complementation for outputting a quantization error, and
subtracting the quantization error from the preceding step from said input signal.
A noise shaping method for feeding back a quantization error component in a quantizer quantizing input signals to an input of said quantizer via a predetermined feedback circuit, comprising the steps of
taking out said quantization error so that the second bit as counted from the most significant bit of input quantization data supplied to said quantizer after complementation and taking out the third and subsequent bits of said input quantization data as such without complementation, said input quantization data being data in which the most significant bit is a sign bit, the next following bit is a unit digit and the following bit or bits are subdecimal digit or digits, and
subtracting the quantization error from the preceding step from said input signals.
A noise shaping method for feeding back a quantization error component in a quantizer quantizing input signals within a range of ±1 into a value of +1 or a value of -1 to an input of said quantizer via a predetermined feedback circuit, comprising the steps of
taking out the second significant bit of quantization input data in a 2's complement representation supplied to said quantizer after complementation and taking out the second significant bit and the following bit or bits of said quantization input data as such without complementation for outputting a quantization error, and
subtracting the quantization error from the preceding step from said input signal.
A noise shaping method for feeding back a quantization error component in a quantizer quantizing input signals within a range of ±1 into a value of +1 or a value of -1 to an input of said quantizer via a predetermined feedback circuit, comprising the steps of
taking out said quantization error so that the second bit as counted from the most significant bit of input quantization data supplied to said quantizer after complementation and taking out the third and subsequent bits of said input quantization data as such without complementation, said input quantization data being data in which the most significant bit is a sign bit, the next following bit is a unit digit and the following bit or bits are subdecimal digit or digits, and
subtracting the quantization error from the preceding step from said input signals.